Evaluating Causal Arctic-Midlatitude Teleconnections in CMIP6

The role of key processes that link amplified Arctic warming and changes in midlatitude weather remains an open question. Observations and model simulations lead to different conclusions, making interpreting Arctic-midlatitude connections difficult. To improve the understanding of these processes, we use a novel method that goes beyond simple correlation analysis, known as causal discovery. This method detects cause–effect relationships from data using graphical models known as causal graphs. In combination with physical reasoning, causal discovery provides a powerful tool to better understand the mechanisms of analyzed teleconnections and to evaluate the performance of climate models. First, we show the causal dependencies in the framework of Arctic-midlatitude teleconnections from observations during 1980–2021. Then we evaluate climate models participating in the Coupled Model Intercomparison Project Phase 6 (CMIP6) by comparing their causal graphs with those derived from observations. This comparison shows if the CMIP6 models reproduce the observed connections in the current climate. We also estimate future changes in Arctic-midlatitude teleconnections toward the end of the century by comparing causal graphs from the CMIP6 historical and Scenario Model Intercomparison Project (ScenarioMIP) simulations. We show that the observations and CMIP6 model simulations do not robustly show Arctic-midlatitude links that arise from Arctic Sea ice variability. The major findings are published in Galytska et al., 2023, https://doi.org/10.1029/2022JD037978.